Bricolage is a term used here and there. Over the Forge, it was some time ago suggested to be the key concept of simulationist play. It was also suggested to be the key concept of all play. I think I agree with the former stance, as I think bricolage is an important part of human thinking.
Bricolage essentially means building something new (or repairing something) with recycled materials. A table is not very stable, so some innocent bundle of newspaper is jammed under one leg. That’s bricolage. Or creating a house system that is an unholy union of Runequest and Spirit of the Century. So: Using stuff with history so that the history remains relevant, though is changed. Term learned from Chris Lehrich’s essay, which is fairly heavy reading (by my standards).
On the elegance of engineering
Suppose I want a roleplaying game that does certain things; for example, is a communal story-creation engine and has a fast and exciting combat system. Maybe something Bourne-like. What can I do? The first option is to build it from scratch; player characters should be fugitive agents, so something that measures if they are about to be caught or get in trouble should be there. They should have some very personal goals to achieve. The natural time limit sets pressure on the goals. (Somebody make this game.) The point is that this is not easy and the end result is likely to not be familiar to the intended audience (the local gaming group, say). This approach can be called engineering. (This is of course also bricolage, but to a lesser degree, or at least of a different kind.)
The second option is to take an existing system that is close to what is wanted and to houserule it (or build a highly derivative system; same thing). The rumours tell that Savage Worlds has pretty fast combat system, so it could probably be hacked into something suitable with minor changes such as tactical renaming of character abilities and tweaking the costs of those, maybe building some new ones or banning unsuitable material. A lot easier and faster; plus, assuming the group is already familiar with Savage Worlds and enjoys it, picking up the new version is easy and likely to end up being fun. This is bricolage, as the end result is heavily defined by the original design of Savage Worlds, which is the relevant history here.
Elegance. Right. I’ll claim that usually an engineered game is more elegant than one constructed via heavy bricolage. A game engineered for specifically this purpose will not have too many irrelevant bits (assuming a good design; adding irrelevant bits is not good design; this is my bias speaking, but I think this is also fairly uncontroversial). The purpose may be very broad; a D&D-like experience with less book-keeping and prep time, for instance, is a totally valid design goal. Heavy bricolage, OTOH, always carries on the assumptions of the original game or games due to retaining their fundamental structure. Often some parts of this fundamental structure are irrelevant to the current game at hand, and hence a source of inelegancy.
So, every game should be engineered to provide a more elegant design
Personally, I don’t think so. Building upon an existing game means a strong foundation and a formidable tradition with answers to several questions that might come up. Watching bricolage in action is something I find fascinating. The end result may be infuriating, though. Witness the parts of your culture you hate the most. Also, those you love the most.
I am prone to always saying that people should design whatever they are doing from the ground up and not mod their favourite game. This is my mistake, as it usually is not true. What people should do is to broaden their toolbox; play and read several different roleplaying games and some other games as a seasoning. The larger set of tools (and places to steal from) allow for better or at least more interesting works. Originality is stealing from sufficiently different sources at the same time.
An important point is that the items one uses for bricolage (games, in this case) will significantly shape the outcome. It follows that using different games as a starting point for design leads to different ways of achieving roughly the same effect. They end results will often feel significantly different, as in the difference between D&D 3rd and Donjon. This is a good thing.